Hi Kevin,
On Mon, 26 Nov 2012 22:24:33 +0100, Kevin Grittner <kgrittn@mail.com>
wrote:
> Marko Tiikkaja wrote:
>> <T2 sees an empty table>
>
> As far as I know you are the first to notice this behavior. Thanks
> for pointing it out.
>
> I will take a look at the issue; I don't know whether it's
> something small I can address in this CF or whether it will need to
> be in the next CF, but I will fix it.
Any news on this front?
>> I'll get back when I manage to get a better grasp of the code.
The code looks relatively straightforward and good to my eyes. It passes
my testing and looks to be changing all the necessary parts of the code.
> Keep in mind that the current behavior of behaving like a regular
> view when the contents are invalid is not what I had in mind, that
> was an accidental effect of commenting out the body of the
> ExecCheckRelationsValid() function right before posting the patch
> because I noticed a regression. When I noticed current behavior, it
> struck me that someone might prefer it to the intended behavior of
> showing an error like this:
>
> ereport(ERROR,
> (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
> errmsg("materialized view \"%s\" has not been populated",
> get_rel_name(rte->relid)),
> errhint("Use the LOAD MATERIALIZED VIEW command.")));
>
> I mention it in case someone wants to argue for silently behaving
> as a regular view when the MV is not populated.
FWIW, I'd prefer an error in this case, but I don't feel strongly about it.
Regards,
Marko Tiikkaja